There is very little written in literature on techniques for estimating the cost of spacecraft and space systems. This is primarily because there is not much information on the cost of space systems freely available for academics to study, and private companies do not want their methods of estimating cost revealed lest their profit margins suffer.
Wertz attempted to change this and started the Journal of Reducing Space Mission Cost in 1998 with the goals of achieving "a better, less expensive, more robust, worldwide space program"[48], and "to bring scholarly, professional research norms and values to this critical problem of space exploration where conversations and reports are frequently dominated by hearsay and unsubstantiated claims"[49]. However, the journal ended in the same year because there were not enough submissions backing up claims of reduced cost with actual data. As Wertz put it "While most people recognized that discussing cost and cost reduction in real and substantive terms was good for the space program as a whole, ultimately, there
were simply not enough organizations willing to take the leap of faith that it would also be good for them. Indeed, it may be true that what is necessary for the industry as a whole is not necessarily good for the individual organizations that discuss it"[50].
The limited literature on cost estimating falls into three main categories:
1. general overviews of cost estimating methods, cost models, and their applications, 2. papers introducing a new cost estimating models or methods,
3. papers discussing the limitations of current cost models and methods.
In the first category, the most notable is the NASA cost estimating handbook which provides NASA’s guidelines for cost estimation and describes the uses of most of the cost models employed by NASA[51]. Wertz and Larson in their book "Reducing Space Mission Cost" have a chapter on cost estimating methods, which primarily focuses on the mathematical methods for developing CERs (Cost Estimating Relationships), and pitfalls to avoid when developing CERs. The rest of the book does not deal with cost estimation but rather best practices to reduce the cost of space missions[52]. Trivailo et al. provides a review of cost estimating models and methods for space mission hardware used by industry and NASA including the types of missions each model is suited for and the phase of the mission where each model is applicable[53].
In the second category Young introduces the method of using optimistic (least), realistic (likely), and pessimistic (most) estimates to define the lower bound, mode and upper bound of a triangle or beta cost distribution[54]. Science Applications International Corporation (SAIC) developed the Microsoft Excel based parametric cost model NAFCOM (NASA/Air Force Cost Model) from historical data on over 155 past NASA and Air Force programs[55]. NAFCOM estimates are primarily based on mass estimates of spacecraft subsystems. The user also has the ability to pick specific historical missions as analogs to the mission they are modeling. NASA’s new cost model, Project Cost Estimating Capability (PCEC) is derived from NAFCOM. The goal of PCEC is to improve upon NAFCOM by
minimizing subjective inputs, and emphasizing quality of input parameters over quantity [56]. Two of the most commonly used commercially provided suits of cost models are the PRICE suit by Price Systems Inc., and SEER by Galorath Inc.[57, 58].
In the third category, Wertz wrote an editorial in the first issue of the Journal of Reducing Space Mission Costthat "Reducing space mission cost is hard if we know what the costs are and virtually impossible if we don’t." Wertz goes on to discuss that since cost data is sensitive it is difficult for cost estimators to gain access to it to produce reliable models. He then asks that governments and companies make their cost data for space programs public or at least allow cost models based on that data to be made public[59]. Jones attributes the high number of missions with high cost overruns to four main factors: undefined, misunderstood or changed mission scope, deliberate low bidding by contractors, excessive optimism of project planners, and finally poor cost estimating methods and data[60]. Keller et al. also admonish current cost estimating methods citing that the average final cost of a US space program exceeds the initial cost estimate by 45%. They blame this partially on parametric cost models claiming that "parametric models only predict the past." Since parametric models are built on historical cost data they will have trouble predicting the cost of any hardware which is significantly different from previous missions[61]. An example of this is the report from the NASA Associate Deputy Administrator for Policy that estimated using NAFCOM that if NASA built the Falcon 9 rocket it would have cost over ten times the amount SpaceX spent. The report also estimated that SpaceX under a firm fixed price contract should have spent about five times as much as they did. The estimators later met with representatives from SpaceX to adjust the inputs to the NAFCOM model and were able to get an updated estimate to match SpaceX’s cost. There were several incorrect assumptions made in the initial estimate which were corrected in the updated estimate.
For example, in the initial estimate the estimators assumed the Merlin engines were new developments when in reality they were nearly identical to the engines flown on the Falcon 1 rocket. The estimators also unintentionally included the weight of the electronics enclosures
in the weight of the electronics. The reason for the gross overestimate of the costs was not necessarily the fault of NAFCOM but rather the estimators not having all the data they needed to complete the estimate properly[62].
The most notable omission from the literature are studies validating cost methods and models. Galorath Inc. and Price Systems Inc., the companies that produce the SEER and PRICE suits of cost modeling software, have conducted their own internal validation studies of their products[63, 64]. However, they do not detail the methods used in the validation studies and the studies have not undergone peer review. The primary focus of these studies appears to be to provide their users with better instructions on how to use their models and to develop standard inputs and assumptions to model various space hardware.
Galorath documents these standard sets of inputs and assumptions in their SEER-H Space Guidance document, an extremely valuable resource to anyone using SEER to estimate the cost of a space system[65]. The absence of independent and peer reviewed validation studies of these models is what motivated the validation studies of the present work found in Section 5.
This section details cost estimation work done as a follow on of the Mars EDLAS (Entry, Descent, and Landing Architecture Study)[19] discussed in Section 2.1. The work in the present section has not been previously published because in 2016, when the work in the present section was completed, the information was considered sensitive. However, this work was presented in a Tech-Forum within NASA Langley’s System Analysis and Concepts Directorate (SACD) and was reviewed by several Subject Matter Experts (SMEs) in both cost estimation and EDL.
The following section of the present section details the use of the parametric cost modeling tool SEER-H to estimate the cost of future space hardware. Sections 3.2, 3.3, and 3.4 discuss the design and mission concept of operations (ConOps) and provide a cost estimate of the Cobra–MRV, ADEPT, and HIAD entry vehicle concepts. Section 3.5 compares the HIAD, ADEPT, and Cobra–MRV concepts, presents an uncertainty analysis of their costs, and provides concluding remarks for the present section.
Note that mass estimates for Cobra–MRV, ADEPT, and HIAD presented in Ta-bles 3.1, 3.2, and 3.3 are slightly different from the masses reported in the EDLAS study and the papers which describe Cobra–MRV, ADEPT, and HIAD. A close reading of the aforementioned papers shows that the masses in the papers do not always agree from doc-ument to docdoc-ument either. This is to be expected at this early stage of the design cycle as designs of the vehicles are rapidly varying. The masses reported in the present work represent the most up to date estimates available at the time this work was completed in late 2016.